Dominique Duncan

Identifying preseizure state in intracranial EEG data using diffusion kernels.

The goal of this study is to identify preseizure changes in intracranial EEG (icEEG). A novel approach based on the recently developed diffusion map framework, which is considered to be one of the leading manifold learning methods, is proposed. Diffusion mapping provides dimensionality reduction of the data as well as pattern recognition that can be used to distinguish different states of the patient, for example, interictal and preseizure. A new algorithm, which is an extension of diffusion maps, is developed to construct coordinates that generate efficient geometric representations of the complex structures in the icEEG data. In addition, this method is adapted to the icEEG data and enables the extraction of the underlying brain activity. The algorithm is tested on icEEG data recorded from several electrode contacts from a patient being evaluated for possible epilepsy surgery at the Yale-New Haven Hospital. Numerical results show that the proposed approach provides a distinction between interictal and preseizure states.

Classification of Alzheimer's disease using unsupervised diffusion component analysis

The goal of this study is automated discrimination between early stage Alzheimers disease (AD) magnetic resonance imaging (MRI) and healthy MRI data. Unsupervised Diffusion Component Analysis, a novel approach based on the diffusion mapping framework, reduces data dimensionality and provides pattern recognition that can be used to distinguish AD brains from healthy brains. The new algorithm constructs coordinates as an extension of diffusion maps and generates efficient geometric representations of the complex structure of the MRI data. The key difference between our method and others used to classify and detect AD early in its course is our nonlinear and local network approach, which overcomes calibration differences among different scanners and centers collecting MRI data and solves the problem of individual variation in brain size and shape. In addition, our algorithm is completely automatic and unsupervised, which could potentially be a useful and practical tool for doctors to help identify AD patients.

Early seizures and temporal lobe trauma predict post-traumatic epilepsy: A longitudinal study

OBJECTIVE Injury severity after traumatic brain injury (TBI) is a well-established risk factor for the development of post-traumatic epilepsy (PTE). However, whether lesion location influences the susceptibility of seizures and development of PTE longitudinally has yet to be defined. We hypothesized that lesion location, specifically in the temporal lobe, would be associated with an increased incidence of both early seizures and PTE. As secondary analysis measures, we assessed the degree of brain atrophy and functional recovery, and performed a between-group analysis, comparing patients who developed PTE with those who did not develop PTE. METHODS We assessed early seizure incidence (n = 90) and longitudinal development of PTE (n = 46) in a prospective convenience sample of patients with moderate-severe TBI. Acutely, patients were monitored with prospective cEEG and a high-resolution Magnetic Resonance Imaging (MRI) scan for lesion location classification. Chronically, patients underwent a high-resolution MRI, clinical assessment, and were longitudinally monitored for development of epilepsy for a minimum of 2 years post-injury. RESULTS Early seizures, occurring within the first week post-injury, occurred in 26.7% of the patients (n = 90). Within the cohort of subjects who had evidence of early seizures (n = 24), 75% had a hemorrhagic temporal lobe injury on admission. For longitudinal analyses (n = 46), 45.7% of patients developed PTE within a minimum of 2 years post-injury. Within the cohort of subjects who developed PTE (n = 21), 85.7% had a hemorrhagic temporal lobe injury on admission and 38.1% had early (convulsive or non-convulsive) seizures on cEEG monitoring during their acute ICU stay. In a between-group analysis, patients with PTE (n = 21) were more likely than patients who did not develop PTE (n = 25) to have a hemorrhagic temporal lobe injury (p < 0.001), worse functional recovery (p = 0.003), and greater temporal lobe atrophy (p = 0.029). CONCLUSION Our results indicate that in a cohort of patients with a moderate-severe TBI, 1) lesion location specificity (e.g. the temporal lobe) is related to both a high incidence of early seizures and longitudinal development of PTE, 2) early seizures, whether convulsive or non-convulsive in nature, are associated with an increased risk for PTE development, and 3) patients who develop PTE have greater chronic temporal lobe atrophy and worse functional outcomes, compared to those who do not develop PTE, despite matched injury severity characteristics. This study provides the foundation for a future prospective study focused on elucidating the mechanisms and risk factors for epileptogenesis.

NIVR: Neuro imaging in virtual reality

Visualization is a critical component of neuroimaging, and how to best view data that is naturally three dimensional is a long standing question in neuroscience. Many approaches, programs, and techniques have been developed specifically for neuroimaging. However, exploration of 3D information through a 2D screen is inherently limited. Many neuroscientific researchers hope that with the recent commercialization and popularization of VR, it can offer the next-step in data visualization and exploration. Neuro Imaging in Virtual Reality (NIVR), is a visualization suite that employs various immersive visualizations to represent neuroimaging information in VR. Some established techniques, such as raymarching volume visualization, are paired with newer techniques, such as near-field rendering, to provide a broad basis of how we can leverage VR to improve visualization and navigation of neuroimaging data. Several of the neuroscientific visualization approaches presented are, to our knowledge, the first of their kind. NIVR offers not only an exploration of neuroscientific data visualization, but also a tool to expose and educate the public regarding recent advancements in the field of neuroimaging. By providing an engaging experience to explore new techniques and discoveries in neuroimaging, we hope to spark scientific interest through a broad audience. Furthermore, neuroimaging offers deep and expansive datasets; a single scan can involve several gigabytes of information. Visualization and exploration of this type of information can be challenging, and real-time exploration of this information in VR even more so. NIVR explores pathways which make this possible, and offers preliminary stereo visualizations of these types of massive data.

Regional and network relationship in the intracranial EEG second spectrum

OBJECTIVE We examined low-frequency amplitude modulations of band power time-series, i.e. the second spectrum, of the intracranial EEG (icEEG) for evidence of support for spatial relationships between different parts of the brain and within the default mode network (DMN). METHODS We estimated magnitude-squared coherence (MSC) of the running power in the delta, theta, alpha, beta, and gamma frequency bands for one-hour background icEEG epochs recorded from 9 patients. We isolated two test areas within the DMN and one control area outside it. We tested if the relationship between DMN areas was stronger than the relationship between each of these areas and the control location, and between all intrahemispheric contact pairs with similar intercontact distance. RESULTS We observed very low values of second spectrum relationship between different parts of the brain, except at very short distances. These relationships are strongest in the delta band and decrease with increasing frequency, with gamma band relationships being the weakest. Our DMN-specific analysis showed no enhanced connectivity in the second spectrum in DMN locations in any frequency band. CONCLUSIONS Though we observed significantly nonzero relationships in lower frequency bands, second spectrum relationships are consistently very low across the entire brain in every frequency band. SIGNIFICANCE This study suggests a lack of support for the DMN in the icEEG second spectrum.

Detecting features of epileptogenesis in EEG after TBI using unsupervised diffusion component analysis

Epilepsy is among the most common serious disabling disorders of the brain, and the global burden of epilepsy exerts a tremendous cost to society. Most people with epilepsy have acquired forms, and the development of antiepileptogenic interventions could potentially prevent or cure these epilepsies [3, 13]. The discovery of potential antiepileptogenic treatments is currently a high research priority. Clinical validation would require a means to identify populations of patients at particular high risk for epilepsy after a potential epileptogenic insult to know when to treat and to document prevention or cure. We investigate the development of post-traumatic epilepsy (PTE) following traumatic brain injury (TBI), because this condition offers the best opportunity to know the time of onset of epileptogenesis in patients. Epileptogenesis is common after TBI, and because much is known about the physical history of PTE, it represents a near-ideal human model in which to study the process of developing seizures. Using scalp and depth EEG recordings for six patients, the goal of our analysis is to find a way to quantitatively detect features in the EEG that could potentially help predict seizure onset post trauma. Unsupervised Diffusion Component Analysis [5], a novel approach based on the diffusion mapping framework [4], reduces data dimensionality and provides pattern recognition that can be used to distinguish different states of the patient, such as seizures and non-seizure spikes in the EEG. This method is also adapted to the data to enable the extraction of the underlying brain activity. Previous work has shown that such techniques can be useful for seizure prediction [6]. Some new results that demonstrate how this algorithm is used to detect spikes in the EEG data as well as other changes over time are shown. This nonlinear and local network approach has been used to determine if the early occurrences of specific electrical features of epileptogenesis, such as interictal epileptiform activity and morphologic changes in spikes and seizures, during the initial week after TBI predicts the development of PTE.

Big data sharing and analysis to advance research in post-traumatic epilepsy

We describe the infrastructure and functionality for a centralized preclinical and clinical data repository and analytic platform to support importing heterogeneous multi-modal data, automatically and manually linking data across modalities and sites, and searching content. We have developed and applied innovative image and electrophysiology processing methods to identify candidate biomarkers from MRI, EEG, and multi-modal data. Based on heterogeneous biomarkers, we present novel analytic tools designed to study epileptogenesis in animal model and human with the goal of tracking the probability of developing epilepsy over time.

The epilepsy bioinformatics study for anti-epileptogenic therapy (EpiBioS4Rx) clinical biomarker: Study design and protocol

The Epilepsy Bioinformatics Study for Anti-epileptogenic Therapy (EpiBioS4Rx) is a longitudinal prospective observational study funded by the National Institute of Health (NIH) to discover and validate observational biomarkers of epileptogenesis after traumatic brain injury (TBI). A multidisciplinary approach has been incorporated to investigate acute electrical, neuroanatomical, and blood biomarkers after TBI that may predict the development of post-traumatic epilepsy (PTE). We plan to enroll 300 moderate-severe TBI patients with a frontal and/or temporal lobe hemorrhagic contusion. Acute evaluation with blood, imaging and electroencephalographic monitoring will be performed and then patients will be tracked for 2 years to determine the incidence of PTE. Validation of selected biomarkers that are discovered in planned animal models will be a principal feature of this work. Specific hypotheses regarding the discovery of biomarkers have been set forth in this study. An international cohort of 13 centers spanning 2 continents will be developed to facilitate this study, and for future interventional studies.

Using Virtual Reality to Improve Performance and User Experience in Manual Correction of MRI Segmentation Errors by Non-experts

Segmentation of MRI scans is a critical part of the workflow process before we can further analyze neuroimaging data. Although there are several automatic tools for segmentation, no segmentation software is perfectly accurate, and manual correction by visually inspecting the segmentation errors is required. The process of correcting these errors is tedious and time-consuming, so we present a novel method of performing this task in a head-mounted virtual reality interactive system with a new software, Virtual Brain Segmenter (VBS). We provide the results of user testing on 30 volunteers to show the benefits of our tool as a more efficient, intuitive, and engaging alternative compared with the current method of correcting segmentation errors.

VRAIN: Virtual reality assisted intervention for neuroimaging

The USC Stevens Neuroimaging and Informatics Institute in the Laboratory of Neuro Imaging (http://loni.usc.edu) has the largest collection/repository of neuroanatomical MRI scans in the world and is at the forefront of both brain imaging and data storage/processing technology. One of our workflow processes involves algorithmic segmentation of MRI scans into labeled anatomical regions (using FreeSurfer, currently the best software for this purpose). This algorithm is imprecise, and users must tediously correct errors manually by using a mouse and keyboard to edit individual MRI slices at a time. We demonstrate preliminary work to improve efficiency of this task by translating it into 3 dimensions and utilizing virtual reality user interfaces to edit multiple slices of data simultaneously.